Intrinsic noise of microRNA-regulated genes and the ceRNA hypothesis

MicroRNAs are small noncoding RNAs that regulate genes post-transciptionally by binding and degrading target eukaryotic mRNAs. We use a quantitative model to study gene regulation by inhibitory microRNAs and compare it to gene regulation by prokaryotic small non-coding RNAs (sRNAs). Our model uses a combination of analytic techniques as well as computational simulations to calculate the mean-expression and noise profiles of genes regulated by both microRNAs and sRNAs. We find that despite very different molecular machinery and modes of action (catalytic vs stoichiometric), the mean expression levels and noise profiles of microRNA-regulated genes are almost identical to genes regulated by prokaryotic sRNAs. This behavior is extremely robust and persists across a wide range of biologically relevant parameters. We extend our model to study crosstalk between multiple mRNAs that are regulated by a single microRNA and show that noise is a sensitive measure of microRNA-mediated interaction between mRNAs. We conclude by discussing possible experimental strategies for uncovering the microRNA-mRNA interactions and testing the competing endogenous RNA (ceRNA) hypothesis.

💡 Research Summary

The paper presents a quantitative study of intrinsic gene‑expression noise in microRNA (miRNA)‑regulated genes and compares it with the noise characteristics of prokaryotic small non‑coding RNAs (sRNAs). The authors construct a stochastic model that captures the essential biochemical steps of both systems. For miRNAs, the model includes catalytic binding (association rate k_on, dissociation rate k_off) followed by multiple rounds of target mRNA degradation with a catalytic turnover rate k_cat. For sRNAs, the model treats binding as stoichiometric: once an sRNA‑mRNA complex forms it is consumed, characterized by a degradation rate k_deg. The master equation describing the joint probability distribution of miRNA, sRNA, and target mRNA copy numbers is solved analytically using linear‑noise approximation (LNA) for mean values and numerically via Gillespie simulations for full noise statistics.

Across a broad range of biologically realistic parameters (miRNA copy numbers 10–10⁴, binding affinities K_D 1–100 nM, degradation rates 0.01–1 min⁻¹), the model predicts that the average steady‑state expression of the target mRNA follows a switch‑like curve as a function of the inhibitor concentration. In the weak‑inhibition regime the mean expression is high, it drops sharply in the intermediate regime, and a residual plateau appears at strong inhibition. Remarkably, this mean‑expression profile is virtually indistinguishable for catalytic miRNAs and stoichiometric sRNAs.

Noise analysis reveals a pronounced peak in the Fano factor and coefficient of variation (CV) at intermediate inhibition strength. This “noise peak” arises because the target mRNA experiences maximal stochastic competition for the limited pool of inhibitors when the system is neither fully repressed nor fully active. The position and amplitude of the peak are essentially the same for miRNA and sRNA systems, and they persist even when key kinetic parameters are varied by an order of magnitude, demonstrating robustness of the phenomenon.

The authors then extend the framework to the competing endogenous RNA (ceRNA) hypothesis, where a single miRNA species regulates multiple distinct mRNA targets that can sequester the miRNA from each other. In this multi‑target model, the mean expression of each mRNA is modestly affected by the presence of the others, but the most striking effect appears in the noise domain. An increase in the abundance of one ceRNA reduces the effective free miRNA concentration, thereby relieving repression on the other ceRNAs and causing a sharp rise in their expression variability. Consequently, intrinsic noise becomes a highly sensitive read‑out of miRNA‑mediated crosstalk, more informative than mean expression alone.

To validate these theoretical predictions, the paper proposes experimental strategies: (i) use fluorescent reporter constructs for several candidate ceRNAs in single‑cell flow‑cytometry or time‑lapse microscopy, (ii) manipulate miRNA levels with CRISPR‑Cas9 knock‑outs or over‑expression vectors to shift the system across the weak, intermediate, and strong inhibition regimes, and (iii) apply RNA‑seq coupled with variance‑stabilizing transformations to quantify genome‑wide noise changes. Bayesian network inference or maximum‑likelihood fitting of the stochastic model to the measured noise data would allow reconstruction of the underlying ceRNA interaction network.

In summary, despite the fundamentally different molecular mechanisms—catalytic turnover for miRNAs versus stoichiometric consumption for sRNAs—the study demonstrates that both systems generate nearly identical mean expression and noise profiles. Moreover, the analysis identifies intrinsic noise as a powerful diagnostic for detecting miRNA‑mediated ceRNA interactions, offering a quantitative framework that can be leveraged to test the ceRNA hypothesis in physiological and disease contexts such as cancer, where dysregulated miRNA‑target networks play a pivotal role.